Gibson’s ecological theory of development is that humans learn in context what objects are things we can interact with. Through interacting with the world and humans around us, we learn what things are tools and levers we can use to move the world. If this is true, what does it mean?

To me, learning to teach kids that the world is malleable in lots of different of ways is of utmost important. You can inspire through poetry, move people with art, code up wonders, edit genomes, etc. I think a cardinal sin of our education system is that rather than teaching kids that knowledge is helping you shape the world, we teach them it’s used to get a letter grade. My very slipshod poor science hypothesis is that people learn that the only way to go anywhere is through good grades which is an illusion the world will quickly prove wrong.

Side-tangent on AI, who could’ve guessed

Looking at the development of AI, we’ve gotten very good at superhuman performance in closed domains. When the rules are set and the objective clear like chess, go, math, or coding, it’s pretty much inevitable that AI is going to surpass human performance. It’s less clear how true this in fields where the rules are extremely murky. How does one win at research? When there aren’t rules or people have to make up the rules to play by, how does it do then?

TODO: Expand this into open-endedness

Sensors

TODO: Flesh out more. I think my point was that sensors in environments are mostly hard-coded right now when they should really be learned. No one told humans that we could augment our senses with microscopes or bits being interpreted as 0s or 1s. How do we get AI that learn new ways of changing the world? Isn’t it kind of amazing that we find new ways of doing that?